Comparative study on hydraulic fracturing using different discrete fracture network modeling: Insight from homogeneous to heterogeneity reservoirs

Conventional discrete fracture network (DFN) modeling is used to simulate hydraulic fracture propagation in fractured low permeability reservoirs. However, natural fractures have a certain degree of roughness, and reservoir rocks have significant heterogeneity characteristics. Therefore, it is necessary to compare the differences in hydraulic fracturing simulation of homogeneous and heterogeneous reservoirs caused by conventional and rough DFN modeling methods. Herein, new-generation algorithms for different DFNs are proposed. Combined the combined finitediscrete element method and different DFNs, four different simulation models are established, covering hydraulic fracturing models from relatively homogeneous to heterogeneous reservoirs. Then, the differences from different DFN modeling methods are compared and discussed. The results show that the relatively homogeneous and heterogeneous models established by conventional DFN may underestimate the bending propagation characteristics and fracturing time and overestimate the proportion of tensile failure. Meanwhile, when the nonuniform distribution of the mechanical parameters of rock blocks is considered, the bending propagation becomes more marked; thus, the fracturing time and the proportion of shear failure are further increased. The above results imply that to accurately predict the hydraulic fracture propagation in a real reservoir, it is vital to consider the roughness of real natural fractures and the nonuniform distribution of mechanical parameters in reservoir hydraulic fracturing modeling.

Automated summary

Conventional discrete fracture network (DFN) modeling is used to simulate hydraulic fracture propagation in fractured low permeability reservoirs. However, natural fractures and heterogeneity in reservoir rocks can significantly impact the simulation results. Therefore, new-generation algorithms for different DFNs are proposed to account for these factors. The study shows that the conventional DFN models may underestimate certain characteristics, such as bending propagation and fracturing time, and overestimate the proportion of tensile failure. Considering the roughness of natural fractures and the nonuniform distribution of mechanical parameters is crucial for accurately predicting hydraulic fracture propagation.